Preliminary results from a multicenter, multi-platform quality control study of the ability of three major vendors' arrays to identify well-characterized diagnostic loci has yielded positive results that could pave the way for wider array usage in cancer cytogenetics.
Jill Hagenkord, a member of the steering committee of the Cancer Cytogenomic Microarray Consortium, told BioArray News this week that the organization recently undertook an evaluation of the performance of cytogenomic arrays manufactured by Affymetrix, Agilent Technologies, and Illumina on multiple sample types; and also examined cross-platform reproducibility of results.
Affy, Agilent, and Illumina were the only vendors who chose to participate in the study, though Hagenkord stressed that the CCMC is "platform neutral" and "works with all vendors."
While the CCMC has not yet reported the final outcome of the study, Hagenkord said that preliminary results indicate that the arrays "were doing their jobs — they were detecting signals when signals were there and not detecting them when they weren’t there."
Hagenkord, who recently was appointed chief medical officer at Complete Genomics, presented the CCMC's preliminary conclusions last month at the Association for Molecular Pathology meeting in Grapevine, Texas.
According to Hagenkord, the CCMC hopes to publish several papers on the study next year, with the hope that the publications will speed the adoption of arrays in clinical cancer cytogenetics, replacing or complementing existing approaches such as fluorescence in situ hybridization or conventional karyotyping.
"The CCMC wants to help usher cancer genomic assays into clinical diagnostics as they mature and clinical utility is established," said Hagenkord. She noted that the organization has a "guideline committee" that is "cross-pollinated" in terms of membership with the American College of Medical Genetics' technical standards and guideline committee.
"The results and data from this study will be automatically fed back into the ACMG committee for their consideration as they continue to refine their standards and guidelines," she said.
The Study
Founded in 2009, the CCMC consists of clinical cytogeneticists, molecular pathologists, and molecular geneticists who are interested in applying microarray technologies to cancer diagnosis and cancer research. According to its website, the CCMC aims to set up platform-neutral standards for cancer microarray designs, create cancer-specific platforms suitable for cancer diagnosis, share cancer microarray data, create a public cancer array database, and carry out multicenter cancer genome research.
There are currently 250 members from 150 academic institutions and commercial entities who subscribe to the CCMC listserv and are regularly invited to participate in CCMC activities, according to the organization.
Last year, the organization designed the QC study, calling it a "Multi-center, cross-platform clinical validation of cancer cytogenomic arrays" (BAN 11/30/2010).
As part of the study, samples are being run on the Affymetrix platform at Creighton and Columbia University; on the Agilent Technologies platform at Baylor College of Medicine and the University of Medicine and Dentistry of New Jersey; and on the Illumina platform at the Children's Hospital of Philadelphia and the Medical University of South Carolina. All of the labs comply with the Clinical Laboratory Improvement Amendments.
Each lab is charged with assessing the ability of the platforms to detect genomic changes in chronic lymphocytic leukemia, myelodysplastic syndromes, and renal cancer in fresh solid and formalin-fixed, paraffin-embedded tumors. The results are then being compared against FISH, conventional karyotyping, and morphology. For the FFPE renal cancer samples, only the Agilent and Affy platforms are being used, Hagenkord noted.
In addition to comparison with traditional assays, the CCMC also seeks to evaluate inter-laboratory reproducibility.
"Our hypothesis is that cytogenomic arrays, regardless of vendor, are all robust, reliable, and reproducible and will all produce the same diagnostic line," said Hagenkord. "Each array may have certain metrics where it outperforms the others, but these metrics will not affect the high-level diagnostic impact of these arrays."
In the CCMC's preliminary analysis of CLL samples, it compared against the standard FISH panel the arrays’ ability to detect certain diagnostic loci — that is genomic changes with known, established clinical utility. The results were presented as a kappa score to assess inter-laboratory reproducibility and concordance across the different platforms, said Hagenkord.
According to Hagenkord, "many clinical assays are plagued by the problem of inter-observer variability, including FISH and metaphase cytogenetics, so typically steps are taken to minimize this, such as averaging the enumeration of multiple different fields by multiple different observers in FISH." In the CCMC's preliminary analysis of cytogenomic arrays, no such efforts were made, she noted.
Each site had one observer, used whichever analysis software it typically uses, and used its own calling thresholds. According to Hagenkord, the kappa scores for the arrays were higher than that reported for FISH. Additionally, all data was sent to an independent analysis team so that the same data files could be processed uniformly, she said.
"We haven’t gotten through all the cohorts yet, but for the CLL arm, we found that the arrays were doing their jobs," she added.
Still, there were some discrepancies between labs. These arose from different laboratories using different analysis settings and thresholds, said Hagenkord. For example, in one CLL case there was a low-level trisomy 12, a variant typically found in 7 percent of CLL cells surveyed by FISH, she said. "Three labs called it, one from each vendor, and three labs did not call it," but the independent analysis team detected it in all datasets, she said.
Subsequent discussions revealed that one of the reporting labs has a 10 percent mosaicism threshold internally, meaning that if the abnormality is present in less than 10 percent of the cells, they do not put it in their diagnostic line. "We realize that our preliminary results, although informative, will show improved kappa scores and concordances when we implement uniform analysis standards," Hagenkord said.
Based on these results, the CCMC reached several preliminary conclusions: that cytogenomic arrays were able to detect clinically actionable genomic changes in cancer; that the kappa scores are comparable to those of FISH, even without multiple reviewers and clear threshold guidelines; and that array-specific analysis guidelines are needed to further improve kappa scores and concordance.
Additionally, the group is recommending SNP arrays for clinical cancer cytogenomics due to their ability to provide internal confirmation of copy number change and detect uniparental disomy.
Hagenkord said that the group hopes to submit a paper in the first half of 2012 concerning analytical sensitivity using the CLL samples, intra-lab reproducibility, and inter-lab and inter-platform reproducibility. Two other manuscripts are expected later next year — one concerning genomic position breakpoint comparison and array-specific analysis recommendations, and the other on FFPE performance and an evaluation of Affy's OncoScan platform.
Jackie Biegel, director of the Cancer Cytogenetics Library at the Children's Hospital of Philadelphia, told BioArray News that it is likely the papers will appear in a supplementary issue of Cancer Genetics, of which Biegel is editor.
'The Ultimate Genetic Test'
In addition to the QC study, the CCMC maintains an array design committee that has recommended coverage of 513 cancer genes plus a genome-wide "backbone" of markers for any cancer cytogenomic array.
The organization has not endorsed any one platform for use, but instead has made a gene list available to all vendors. According to Hagenkord, the CCMC is seeking to refine what constitutes recommended coverage, but for now it is up to the vendors to decide what content they put on the arrays.
"The vendors have been really receptive to the input of the CCMC and we are in frequent communication with them as they design new products," said Hagenkord.
Over the past year, a number of array manufacturers have rolled out products based on the CCMC design. BlueGnome, Signature Genomics, CombiMatrix, Ambry Genetics, and OGT have all debuted or announced plans to make available cancer cytogenomic arrays (BAN 9/13/2011).
In October, Agilent introduced its SurePrint G3 CGH+SNP cancer catalog microarrays, also based on the CCMC design. According to Hagenkord, Agilent's chip targets the 513 genes with "ultra-high coverage and a modest backbone."
Affy and Illumina, meantime, have cross-referenced the CCMC cancer gene list to the coverage on their genome-wide high density arrays: the SNP Array 6.0 and the Infinium HD Human Omni1 BeadChip, respectively, in the case of the QC study. "Since both Illumina and Affy already target 99 percent of all RefSeq genes, they get the CCMC cancer genes almost by default, plus the high-coverage backbone," she said.
The CCMC is also interested in expanding its focus beyond arrays to include clinical cytogenetics performed using next-generation sequencing.
"The CCMC recognizes that data obtained from whole-genome sequencing can be used to recreate a karyotype in silico, just as the cytogenomic arrays are doing today," said Hagenkord. "Moreover, WGS can potentially detect balanced translocations and inversions and somatic sequence mutations in addition to genome-wide copy number and loss of heterozygosity," she said.
Whole-genome sequencing is "the ultimate genetic test," said Hagenkord, "and I think the technology and analysis are maturing more quickly than almost anybody imagined." She added that the CCMC board of directors intends to incorporate sequencing into its current mission, since the technology and cytogenomics are "rapidly merging into one discipline."
Indeed, a keynote speaker at the CCMC's annual meeting in Chicago this past August was Elaine Mardis, professor of genetics and co-director of the Genome Institute at Washington University.
According to a paper containing an overview of the meeting, Mardis' presentation, entitled "Tumor evolution by deep digital
sequencing," provided a "glimpse into the future of cancer diagnostics, when each tumor and matched normal specimen will be analyzed by whole-genome sequencing."
Toward New Standards
At the Chicago meeting, Hutton Kearney, cytogenetics director at Mission Hospitals in Asheville, NC, and Daynna Wolff, director of the cytogenetics lab at the Medical University of South Carolina, reviewed the draft ACMG standards and guidelines for cancer microarrays, which are in the "final stages of development," according to the overview.
Microarrays in neoplasia will be recommended as an adjunct to karyotypes and FISH until "sufficient evidence is acquired to recommend them as a first-line test," according to the overview. Key components of the draft guidelines will also address CCMC's validation of the platform, which are expected to be presented in full by the time of next summer's annual meeting.
The new guidelines are being drafted just a year after ACMG revised its guidelines to recommend that cytogenomic arrays be used as first-tier diagnostics for congenital abnormalities. Mission Hospitals' Kearney was one of the authors of the new guidelines that made arrays the standard of care in the US for clinical constitutional cytogenetics.
Hagenkord said that, going forward, the CCMC plans to "work with vendors and other stakeholders to ensure appropriate technical standards and validation guides, interpretative recommendations, and educational opportunities."
Have topics you'd like to see covered in BioArray News? Contact the editor a1t jpetrone [at] genomeweb [.] com.